JPH08265761A - Image data post-processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To eliminate blocking by referencing an estimate mean pixel value of a current block through the calculation of a difference of pixels of adjacent blocks along a border line of the current block so as to adjust the total pixel values of the current block. SOLUTION: A mean value computer 202 receives IDCT data of a current block from a line L13 and provides an output of a mean value MV of luminance values of pixels of the current block along a border line of adjacent blocks. A mean value estimate device 212 receives pixel values of the adjacent blocks from a line L14 and provides an output of an estimate mean value EMV of the adjacent blocks based on a difference of the pixels of the current block along the border line of the current block. A multiplier 204 receives a DC quantized step size QS of the current block via a line L6 and calculates a value QS/2. In the case that a difference between the mean value MV and the estimate mean value EMV is within the QS/2, all the pixel values of the current block are corrected by the difference between the mean value MV and the estimate mean value EMV, and in the case that the difference is more than the QS/2, the pixel values of the current block are corrected by the QS/2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a post-processing method used in a video signal decoding device, and more particularly, to post-processing the decoded video data effectively to obtain the decoded video data. The present invention relates to a post-processing method capable of removing a blocking phenomenon appearing at a block boundary.

[0002]

BACKGROUND OF THE INVENTION In various electronic / electrical applications such as high definition television (HDTV) and videophone systems,
Generally, it is necessary to transmit the video signal in digital form. When this video signal is represented in digital form, a large amount of digital data is generated. However, due to the limited available frequency bandwidth of conventional transmission channels, the amount of data to be transmitted must be compressed in order to transmit large amounts of digital data over the limited channel. Among various video signal compression techniques, it is known that a hybrid coding technique that combines a so-called statistical coding method and a temporal and spatial compression method is most effective.

Most hybrid coding techniques rely on adaptive inter / intra mode coding.
g), orthogonal transform, quantization of transform coefficient, RLC (run length coding) and VLC (variable length coding) are used. Adaptive inter / intra mode coding selects a video signal for subsequent orthogonal transformation from PCM (pulse code modulation) data of the current frame or, for example, DPCM (differential pulse code modulation) data based on the variance of PCM data. Is a process. Inter-mode coding, which is known as a prediction method, is based on the concept of reducing redundancy between adjacent frames, in which motion of an object between the current frame and one or two adjacent frames is determined. , The process of predicting the current frame by the moving flow of the object and generating an error signal representing the difference between the current frame and its predicted value. Such an encoding method is described, for example, in the article by Staffan Ericsson, "" Fixed an
d Adaptive Predictors for Hybrid Predictive / Transf
orm Coding '', IEEE Transactions on Communications,
COM-33, No. 12, 1291-1301 (December 1985) ”, Ninomi
The paper by ya and Ohtsuka ““ A Motion-Compensated Inter
frame Coding Scheme for Television Pictures ", IEE
E Transactions on Communications, COM-30, No.1,201-2
P. 10 (January 1982) ".

Orthogonal transforms use the spatial correlation between video data such as PCM data of the current frame or motion-compensated DPCM data to reduce or remove spatial redundancy between them and to digitally Convert blocks of video data into a set of transform coefficients. Such techniques are described, for example, in the article by Chen and Pratt, "" Scene Adaptive Code.
r '', IEEE Transactions on Communications, COM-32, N
o.3, pages 225-232 (March 1984) ”. Quantization of such transform coefficient data, zigzag scanning, RL
By processing through C and VLC, the amount of data to be transmitted can be effectively compressed.

The coded video data is transmitted to the video signal decoder in the video signal decoding device of the receiver through a normal transmission channel. In this video signal decoder, the original video data is reconstructed by performing the reverse process of the encoding operation. However, in the reconstructed video data or the decoded video data, an artifact called a blocking phenomenon in which the border line of the block becomes noticeable at the receiver is displayed. Such a blocking phenomenon occurs frequently because one frame is encoded in block units,
Furthermore, if the quantization step size is large (ie,
Becomes more noticeable when the block is quantized more coarsely). Specifically, if any block is brighter or darker than its neighbors, and if a large quantization step size is used to quantize the intra block DC coefficients, then that block and its neighbors The difference in strength with the block becomes even more apparent. As a result, the blocking phenomenon becomes more severe, and the image quality of the image deteriorates.

Therefore, various types of post-processing techniques have been proposed to improve the quality of decoded video data by reducing the blocking phenomenon. In general, these post-processing techniques use post-processing filters to further process the decoded video data.

In the post-processing technique using the post-processing filter, the decoded video data is filtered pixel by pixel using a low pass filter having a predetermined cutoff frequency. A post-processing technique that employs a post-processing filter is disclosed in co-pending US patent application Ser.
/ 431,880 specification, `` IMPROVED POST-PROCESSING MET
It is disclosed under the name "HOD FOR USE IN AN IMAGE SIGNAL DECODING SYSTEM". This technique can reduce the blocking phenomenon appearing at the block boundary of the decoded video data by repeatedly post-processing each filtered video data. However, since such a post-processing technique filters the decoded video data without considering the position of each pixel included in the block, the blocking phenomenon along the block boundary cannot be sufficiently removed.

[0008]

Therefore, the main object of the present invention is to use in a video signal decoding device, by adjusting or matching the pixel value of the current block and the pixel value of its adjacent block. The object of the present invention is to provide a post-processing method capable of sufficiently removing a blocking phenomenon appearing at a block boundary in decoded video data.

[0009]

In order to achieve the above object, according to the present invention, video data used in a video signal decoding apparatus for decoding an encoded digital video signal and for post-processing the current block is used. In the post-processing method, the encoded digital video signal has a plurality of encoded blocks made up of quantized transform coefficients, and the block corresponding to any one of the encoded blocks is: It has one quantized DC coefficient quantized based on a DC quantization step size and a predetermined number of quantized AC coefficients, and a boundary line between the current block and its adjacent block. A first step of obtaining an average current pixel value (ACV) of the pixels of the current block located along the line, and an average pixel value of the pixels of the current block using the variation of the pixel values of the pixels in the adjacent block. By estimating a second step of generating an estimated current pixel value (ECV), the average current pixel value (ACV), the estimated current pixel value (ECV) and the DC
A third step of adjusting a pixel value of a pixel in the current block according to a quantization step size.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 shows a schematic block diagram of a video signal decoding apparatus according to the present invention capable of removing a blocking phenomenon. As shown, the encoded video bitstream is transferred from a conventional encoder (not shown) to a buffer 10 in the decoding device.
Supplied to zero. The buffer 100 receives an encoded video bit stream and supplies it to a variable length decoder (VLD) 102 at a constant bit rate.

In the VLD 102, first, an input coded video bit stream is demultiplexed, and an intra / inter mode signal, coded video data (that is, a plurality of coded blocks made up of quantized transform coefficients) and a variable length. , And then decoding the quantized transform coefficients of the plurality of encoded blocks to obtain blocks of a plurality of quantized transform coefficients Qfs and a set of quantization step sizes for each of these blocks. Is supplied to the dequantizer 106 via the line L10, the coded motion vector of the variable length is also decoded, and supplied to the motion compensator 128 via the line L12. Qfs for each block (eg 8 × 8 block) is one DC coefficient and 63 ACs.
Has a coefficient, and one set of quantization step sizes has two kinds of quantization step sizes, one of which represents a quantization step size for a DC coefficient, and the other one of which is AC.
Indicates the quantization step size for the coefficient.

In the inverse quantizer 106, the line L4
The Qfs of one block is converted into a set of Dfs using the corresponding set of quantization step sizes determined by the intra / intermode signal supplied from the VLD 102 via
Convert to CT coefficient. Then, the DCT coefficient of this set is I
The DCT (discrete inverse cosine transformer) 110 is supplied, and the quantization step size for the DC coefficient is supplied to the post-processing block 124. IDCT110 is DCT
The coefficient set is inversely transformed into one block of IDCT data, and this data is supplied to the adder 116.

On the other hand, the motion compensator 128 includes a VLD 102.
From the previous frame stored in the frame memory 126, a block of pixel data corresponding to a block of IDCT data is extracted via a line L18 using the motion vector supplied from The compensation data is supplied to the switch SW10. In this switch SW10, when the inter-mode signal is input through the line L14, the motion compensation data is sent to the adder 116 and is added to the IDCT data to supply the decoded data signal to the post-processing block 124. When the intra mode signal is input via the line L4, the switch SW10 is closed, and the IDCT data from the IDCT 110 is supplied to the post-processing block 124 via the adder 116 without any addition operation. .

In the post-processing block 124, a so-called blocking phenomenon caused by a large difference in the luminance value of the pixel between the current block and its adjacent block is removed as follows. That is, the blocking phenomenon may occur in the adjacent block after first calculating the average luminance value (MV) of the pixels of the current block located along the boundary line between the current block and the adjacent block. The estimated average value (EMV) of the pixels in the current block is obtained by estimating the average luminance value of the pixels in the current block using the pixel value variation of the pixel, and then MV and EMV are compared with each other, Based on the result and the DC quantization step size (QS) of the current block, it can be removed by adjusting all pixel values of the current block.ー QS
When / 2 <(MV-EMV) <QS / 2, all pixel values of the current block are increased by (MV-EMV), and (M
If V−EMV) ≧ QS / 2, all pixel values of the current block are reduced by QS / 2, and (MV−EMV) ≦ −Q
In the case of S / 2, all pixel values of the current block are increased by QS / 2. Regarding this post-processing block 124,
This will be described in detail below with reference to FIG.

The post-processing block 124 includes an average value calculator 2
02 and a mean value estimator 212, a multiplier 204, a first
And second adders 206 and 220, first and second subtractors 208 and 218, and first and second selectors 210.
And 214 and a buffer 216.
The IDCT data is input from the adder 116 to the average value calculator 202 and the buffer 216 via the line L13. The average value calculator 202 stores the pixel value of the current block and calculates the first average value MV. As shown in FIG. 3B which is an enlarged view of FIG. 3A, for example, 8 × 8
The MV for a block is calculated by the following equation (1).

[0016]

[Equation 1] Here, C (i, j) represents the luminance value of the pixel located at the i-th row and the j-th column in the current block. As can be seen from the above equation (1), MV is the average value of the luminance values of the pixels of the current block located along the boundary line between the current block and its adjacent block.

On the other hand, all pixel values of adjacent blocks placed on the upper side and the left side of the current block in the figure are input from the frame memory 126 to the average value estimator 212 via the line L14. The average value calculator 212 calculates the average value EMV. Referring to FIG. 3B, when adjacent blocks of the current block are located on the upper side and the left side, two 8 × 8
The EMV for a block is calculated by the following equation (2).

[0018]

[Equation 2] Here, U (i, j) is the luminance value of the pixel in the i-th row and the j-th column in the adjacent block located on the upper side, and is L
(I, j) is i in the adjacent block located on the left side
The luminance values of the pixels in the th row and the j th column are respectively represented. As can be seen from the above equation (2), the EMV is obtained by using the change in the pixel value of the pixel in the adjacent block.

To adjust the pixel value of the current block based on the difference between the mean value MV and the estimated mean value EMV, the multiplier 204 adds 1 to the QS from the dequantizer 106.
// 2, and the multiplication result QS / 2 is supplied to the first adder 206 via the line L24. First adder 206
Adds the MV from the average value calculator 202 and the QS / 2 from the multiplier 204, and the addition result MV + QS / 2
Are supplied to the first selector 210 via the line 26. The first selector 210 selects the smaller value of the MV + QS / 2 from the adder 206 and the EMV from the average value estimator 212 to select the smaller value and outputs the second selector 214.
Supply to The first subtractor 208 calculates the difference between MV and QS / 2, and supplies this subtraction result MV-QS / 2 to the second selector 214. The second selector 214 includes the MV-QS / 2 from the subtractor 208 and the first selector 21.
Of the selected value from 0, the larger value is selected and supplied to the second subtractor 218. The second subtractor 218 has a second
Subtracting MV from the value from selector 214 produces one of QS / 2, MV-EMV or -QS / 2. In the second adder 220, the second
The adjusted pixel value of the current block is generated by adding the result from the subtractor 218 and all pixel values of the current block delayed in the buffer 216. The adjusted pixel value is simultaneously supplied to the frame memory 126 and the display device.

While specific embodiments of the invention have been described above, it will be apparent to those skilled in the art that various modifications can be made without departing from the scope of the claims set forth herein.

[0021]

Therefore, according to the present invention, the pixel value of the current block is adjusted by using the average value of the pixels of the current block and the average value of the adjacent blocks, and the block of the decoded video data is adjusted. By significantly reducing the blocking phenomenon that appears at the boundary, the image quality of the video signal can be improved.

[Brief description of drawings]

FIG. 1 is a schematic block diagram of a video signal decoding apparatus according to the present invention.

FIG. 2 is a detailed block diagram of the post-processing block of FIG.

3A is an explanatory diagram showing a current block and its adjacent blocks in a frame, and FIG. 3B is a diagram showing FIG. 3A.
FIG. 9 is an enlarged view of FIG. 9B, which is a diagram for explaining an operation of removing a blocking phenomenon by adjusting pixel values located at boundaries between blocks.

[Explanation of symbols]

 100, 216 Buffer 102 VLD (Variable Length Decoder) 106 Inverse Quantizer 110 IDCT (Discrete Inverse Cosine Transform) 116, 206, 220 Adder 124 Post Processing Block 126 Frame Memory 128 Motion Compensator 202 Average Value Calculation Unit 204 Multiplier 208, 218 Subtractor 210, 214 First and second selector 212 Average value estimator SW10 switch

Claims (2)

[Claims] 1. A video data post-processing method used in a video signal decoding apparatus for decoding an encoded digital video signal, for post-processing a current block, wherein said encoded digital video signal is quantized. A plurality of coded blocks of the transformed coefficients, and the current block corresponding to any one of the coded blocks is quantized based on a DC quantization step size (Q _s ). It has one quantized DC coefficient and a predetermined number of quantized AC coefficients, and the average current of the pixels of the current block located along the boundary line between the current block and its adjacent block. Pixel value (A
CV) and a variation of pixel values of pixels in the adjacent block to generate an estimated current pixel value (ECV) by estimating an average pixel value of pixels in the current block. The second step of performing, the average current pixel value (ACV), the estimated current pixel value (EC
V) and a third step of adjusting a pixel value of a pixel in the current block according to the DC quantization step size. 2. The third step is to calculate the estimated current pixel value (E) from the average current pixel value (ACV).
CV) is subtracted, and if (ACV-ECV) ≧ QS / 2, all pixel values of the current block are decreased by QS / 2, and (ACV-ECV)
If V) ≤ -QS / 2, increase all pixel values of the current block by QS / 2, and -QS / 2 <(ACV-EC
The method according to claim 1, further comprising the step of increasing all pixel values of the current block by (ACV-ECV) when V) <QS / 2.